Two comments and two questions.

Q1:  XSLT patterns don't have the zero-length assertion \b available
to match a word boundary. This may result in unexpected matches. With
analyze-string it is not possible to apply the usual trick of adding
an extra character before and after the string. So how can an exact
match be done here?

Q2: If the index or document is big, it might be faster to have
xsl:key on the indexTerms. Is it possible to construct such a key with
the matching string being the original<term/>  content *in lowercase*?

not sure what you mean but
<xsl:key match="something" use="lower-case(term)"/>
might be the answer.

Can it be done by constructing a temporary tree and applying xsl:key
to that?

-W

________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.